Siren Sound (Alert?) - All Temps Good; No Other Indicators (SOLVED)

[_0m0t3ur]

I have heard an audible alert of some kind, three separate instances.  It sounds like a siren (nah, nuh, nah, nuh) and lasts between 6-8 seconds.  I heard it twice yesterday, and once today.

 

Supermicro X8SIL-F

Xeon X3460

4 x 4 GB Nanya 1333 ECC RAM

LSI 2911-8i HBA

 

My server has been running for two days.

 

I just finished my server and installed unRAID to a Sandisk Micro Cruzer.  It booted fine.  It's been running fine.

 

I have been using the Preclear plug-in to manually preclear my five 4TB WD Reds.  I started with just one.  After it completed the initial read, I started the pre-clear on a second drive.

 

Temps:

 

CPU:  32 C

Motherboard:  33 C

HDD 1:  29 C

HDD 2:  29 C

HDD 3:  30 C

HDD 4:  29 C

HDD 5:  30 C

 

No SMART drive errors. 

Average CPU load while preclearing two drives between 30%-50%.

I ran MemTest prior to booting unRAID, with no issues.

 

I looked at the motherboard manual but they talk about beeps during post.  Thank in advance.  Any help would be appreciated.

 

Why am I hearing this kind of alert?

 

Update 06.09.2018 9:00 pm:

 

After reapplying thermal paste to my heat sink/CPU, I haven't had any "British" siren alerts.  For now, it appears to be solved.

 

Update 06.10.2018 2:20 pm:

 

Not solved, I guess.  10 minutes ago I just got my first British siren audible alert since I reapplied thermal paste to my heatsink/CPU.  At this point, I'm not sure.  I have read that others in this forum have my same board (Supermicro X8SIL-F) enabled the jumper to silence audibles.  But that doesn't seem ideal.

 

Update 06.10.2018 3:11 pm:

 

The alert went off again.  I noticed that the fan speed displayed at that moment for fan header 1 was 78 RPM.  This definitely looks like the culprit, causing the alarm.

 

Update 06.14.2018 4:30 pm:

 

Disabled the onboard speaker.

Edited June 14, 2018 by DoItMyselfToo
Updated thread.

[pwm]

Probably a read error of a fan RPM or a temperature. It is quite common that people have problems with SMBus (basically I2C) readings. Some motherboards are worse than others. And some software are way worse than others when it comes to filtering the measurements and not jumping into instant panic for a single incorrect reading.

[_0m0t3ur]

2 minutes ago, pwm said:

Probably a read error of a fan RPM or a temperature. It is quite common that people have problems with SMBus (basically I2C) readings. Some motherboards are worse than others. And some software are way worse than others when it comes to filtering the measurements and not jumping into instant panic for a single incorrect reading.

 

Are you saying that I might be having a high temp or low RPM that is not being reported to unRAID, but the motherboard is alerting?

[pwm]

Just now, DoItMyselfToo said:

 

Are you saying that I might be having a high temp or low RPM that is not being reported to unRAID, but the motherboard is alerting?

 

Yes, I assume it's the motherboard itself that is activating the alarm based on an incorrect reading. And then 5-10 seconds later when the motherboard makes a new reading it gets a correct value and then directly mutes the alarm again. It might be meaningful to google to see if other people have issues with your motherboard. Or if there are a newer BIOS.

[_0m0t3ur]

I initially searched unRAID and didn't see anything.  Then I posted here.  Then I Googled the issue and found an unRAID forum post from 2011.  I PM'd the OP.  Hopefully there is a solution.

 

Oh, yeah, I have the latest BIOS and IPMI.

 

Edited June 6, 2018 by DoItMyselfToo

[pwm]

This is another tread about people having issues with bad readings on the SMBus

 

[_0m0t3ur]

Thank you.  I'll look at that.

[_0m0t3ur]

First off, I finished pre-clearing two 4TB drives using the Preclear plug-in.  During that time, I heard a total of 6 siren sounds, lasting 6-8 seconds.  The CPU, as I mentioned in my first post, was average loading between 30-50% with two drives.  The OS (unRAID) was showing CPU temps in the low 30's C and motherboard temp in the mid 30"s.  There are several options

 

After reading PWM's posts in this thread and some searching on the Internet, I have come to the conclusion that the siren sound is likely a momentary high temp reading.  So, I rebooted into MemTest86 and ran it in both default mode and CPU multi-thread mode.  I was seeing CPU temps in the 45-50 C range for the default mode; and for the multi-thread mode the CPU was ranging between 52-59 C.

 

I was out of thermal paste, so I ordered Gelid GC-Extreme because I could get it with free overnight pick-up.  When I removed the Arctic Freezer 12 CPU cooler, while there was ok coverage, the cooler released very easily, which confirmed for me that there was too little thermal paste from my initial application.  I cleaned the CPU and cooler.  Then reapplied the paste and secured.  The cooler felt tighter to the board/CPU than my initial installation.

 

I rebooted into MemTest86, running in multi-thread mode, getting lower temps in the 45-51 C range.

 

Additionally, during this process, I discovered I was losing time, so I replaced my CMOS battery (CR2032).  Lost time issue is corrected.

 

We'll see about the siren sounds.  I'll update.

Edited June 8, 2018 by DoItMyselfToo

[strike]

Do you have anything other than the HBA? Specifically in the closest pcie slot? I tried to put a GPU right next to my Raid card once, that didn't go well. After an hour or so the alarm went off so I had to change the slot for my GPU. My Raid card was hot as hell, nearly burned myself. 

[_0m0t3ur]

Do you have anything other than the HBA? Specifically in the closest pcie slot? I tried to put a GPU right next to my Raid card once, that didn't go well. After an hour or so the alarm went off so I had to change the slot for my GPU. My Raid card was hot as hell, nearly burned myself. 

No. I have only the HBA. The northbridge gets very warm too. I wish they came with a fan.


Sent from my iPhone using Tapatalk

[_0m0t3ur]

After more searching of this issue, I decided to disable the onboard speaker, because I believe that the audible alert is not representative of a real problem in the system.  All my temps are fine and the one time I saw a low RPM reading of 78, my fans were all spun up on high.  I've measured temps on board using a laser thermometer, getting measurements that are fairly close to the readings reported by the Dynamix System Temperature plugin.  Others with this MB have reported disabling the onboard speaker after experiencing this same problem.  One in particular told me that he ran his server after for six years with no issues.

 

I suspected possible momentary temperature spikes or RPM changes.  I saw one instance of fan header 4 showing 0 RPM and alarming, even though fan header 4 is not being used, so there would be no tach or pulse modulation as a basis for an alert on that header.  As I mentioned elsewhere in this thread, I think there was too little paste, so I replaced the cooler/CPU thermal paste, hoping to insure even heat dissipation.  The only other really warm area on the board is the northbridge heat sink.  I'm looking at adding a small fan to the heat sink to see if it will keep the temps low, though from what I've read, the temps are not really too high to begin with.

Edited June 14, 2018 by DoItMyselfToo

[pwm]

I'm pretty sure it's read errors - i.e. problem with the communications with the supervisor chips that measures the temperatures/rpm/voltages.